Piece Resolution : Towards Larger

This paper focuses on two aspects of piece resolution in backward chaining for conceptual graph rules 13]. First, as conceptual graphs admit a rst-order logic interpretation, inferences can be proven by classical theorem provers. Nevertheless, they do not use the notion of piece, which is a graph notion. So we deene piece resolution over a class of rst-order logical formulae: the logical rules. An implementation of this procedure has been done, and we compare the results with the classical SLD-resolution (i.e. Prolog). We point out several interesting results: it appears that the number of backtracks is strongly reduced. Second, we point out the similarities between these rules and database data dependencies. The implication problem for dependencies is to decide whether a given dependency is logically implied by a given set of dependencies. A proof procedure for the implication problem, called \chase", was already studied. The chase is a bottom-up procedure: from hypothesis to conclusion. This paper introduces a new proof procedure which is top-down: from conclusion to hypothesis. Indeed, we show that the implication problem for dependencies can be reduced to the existence of a piece resolution.